Utilisation of genetic variation by marker assisted selection in commercial dairy cattle populations

Utilisation of genetic variation by marker assisted selection in commercial dairy cattle populations

Livestock Production Science 59 (1999) 51–60 Utilisation of genetic variation by marker assisted selection in commercial dairy cattle populations R.J...

142KB Sizes 20 Downloads 45 Views

Livestock Production Science 59 (1999) 51–60

Utilisation of genetic variation by marker assisted selection in commercial dairy cattle populations R.J. Spelman b

1 ,a ,

*, D.J. Garrick a , J.A.M. van Arendonk b

a Institute of Veterinary, Animal and Biomedical Sciences, Massey University, Palmerston North, New Zealand Animal Breeding and Genetics group, Wageningen Institute of Animal Sciences, Wageningen Agricultural University, Wageningen, The Netherlands

Received 12 May 1998; accepted 16 November 1998

Abstract Potential genetic benefits of marker assisted selection (MAS) were evaluated by calculating selection response resulting from four pathways of selection. Genetic variation was partitioned into polygenic and loci that were in linkage disequilibrium with marker loci or haplotypes. The percentage of genetic variation that was marked was varied from 0 to 100%. These assumptions describe the degree of genetic knowledge that may be available in 10 years. Three breeding strategies with markers were evaluated: progeny test scheme (BMARK); progeny test scheme but unproven bulls allowed on the bull to bull selection path (YBULL); and a breeding programme where cows without lactation information and bulls without progeny information were eligible for selection (OPEN). Rates of genetic gain (per year) with no marked genetic variance were 0.26 s G for the BMARK and YBULL schemes and 0.28 s G for the OPEN scheme. On average, an increase of 1% marked genetic variance resulted in an increase in genetic gain of | 0.25% for the BMARK scheme, 0.5% for the YBULL scheme and 1% for the OPEN scheme. Maximum genetic response (100% marked genetic variance) for the BMARK scheme was 1.24 times that achieved with no marked genetic variance, 1.52 times for the YBULL scheme, and 2.05 times for the OPEN scheme. Changes in the structure of the breeding scheme are needed to fully gain the benefits of identified loci especially for medium to large proportions of marked genetic variance.  1999 Elsevier Science B.V. All rights reserved. Keywords: Marker assisted selection; Genetic variation; Dairy cattle

1. Introduction Quantitative trait loci (QTL) experiments in dairy *Corresponding author. Livestock Improvement, Private Bag 3016, Hamilton, New Zealand. Tel.: 164-6-7856-0642; fax 1646-7858-2741. E-mail address: [email protected] (R.J. Spelman) 1 On leave from Livestock Improvement Corporation, New Zealand.

cattle using granddaughter and daughter designs are successfully detecting QTL (Georges et al., 1995). However, the proportion of genetic variation for individual traits that has been explained to date in these experiments is usually less than 15%. For milk production traits with heritabilities of some 30%, this is equivalent to | 5% of the phenotypic variance. The expected genetic improvement from marker assisted selection is some 2–10% assuming up to 15% genetic variance is identified (Spelman and

0301-6226 / 99 / $ – see front matter  1999 Elsevier Science B.V. All rights reserved. PII: S0301-6226( 99 )00003-2

52

R. J. Spelman et al. / Livestock Production Science 59 (1999) 51 – 60

Garrick, 1997, 1998; Mackinnon and Georges, 1998). Experimental techniques such as selective DNA pooling (Darvasi and Soller, 1994) applied to large half-sib families, that exist in commercial dairy populations, have the power to explain more genetic variation than granddaughter and daughter designs with less genotyping effort (Spelman et al., 1998). The potential of selective DNA pooling has been demonstrated in the Israeli dairy population with a large proportion (0.5–0.75) of genetic variation for one milk production trait being explained (Lipkin et al., 1998; M. Soller personal communication). A limitation for MAS is that linkage phase has to be estimated for each family and confirmed in subsequent generations. To overcome this problem, the QTL themselves would have to be identified or markers or marker haplotypes identified that are in linkage disequilibrium with the QTL (Smith and Smith, 1993). Linkage disequilibrium mapping has been applied successfully to identify single genes in livestock (Charlier et al., 1996) and its application to complex traits in humans is viewed positively (Risch and Merikangas, 1996). However, Baret and Hill (1997) outline that the application of linkage disequilibrium mapping in livestock is limited. Linkage disequilibrium mapping is being currently applied to complex traits in livestock (M. Georges, personal communication) and in 10 years time there is the possibility that in dairy cattle populations a large proportion of the marked genetic variance will be in linkage disequilibrium with marker loci (haplotypes).

Smith (1967) and Lande and Thompson (1990) among others have studied genetic responses to a single generation of marker assisted selection with the assumption of genes in disequilibrium. However, to date the utilization of disequilibrium by MAS in a dynamic cattle breeding scheme has not been investigated. The objective of this study is to identify the possible genetic responses that could be achieved with MAS assuming a large proportion of the genetic variation is in disequilibrium with markers. The study investigates the gains from MAS that can be made in a dynamic breeding scheme where the age at selection is not fixed.

2. Method

2.1. Simulation model A deterministic simulation model accounting for four pathways of selection and overlapping generations was developed. Population parameters were based on the New Zealand Holstein-Friesian breed and are outlined in Table 1. The base breeding scheme comprises 140 bulls that are progeny tested on 85 daughters each and receive their progeny test proof at the age of 5 years. To be eligible for selection in the base scheme a female had to be at least 1 year of age and a male had to have 85 daughters with at least one complete

Table 1 Population parameters used to calculate annual genetic gain Cow population Age structure; 22% calves, 16% 1 year olds (yos), 14% 2 yos, 13% 3 yos, 11% 4 yos, 9% 5 yos, 8% 6 yos and 7% 7 yos 272 000 calves to 7 yo cows eligible for selection as bull dams 450 selected each year as bull dams No selection on the cow to cow pathway Bull population 140 bulls progeny tested per year 2% death rate (three bulls) per year Receive progeny test proof as 5 year olds on 85 daughters Ten bulls selected for bull to cow pathway each year from live 5, 6 and 7 yos Three bulls selected for bull to bull pathway each year from 5 and 6 yos (dead bulls eligible for selection as frozen semen held)

R. J. Spelman et al. / Livestock Production Science 59 (1999) 51 – 60

lactation record. Ten bulls were selected from the live 5-, 6- and 7-year-old bulls for the bull to cow path and three bulls for the bull to bull path from live or dead (frozen semen stored) 5- and 6-year-old bulls. The genetic contribution of young bull inseminations undertaken in the cow population for progeny testing was ignored. Four hundred and fifty lactating cows were selected from a population of some 272 000 females (calves through to 7-year-old calves eligible for selection) for the cow to bull path. No selection was undertaken on the cow to cow path; females were produced from the 30% of 1 year olds that were artificially inseminated and from all older age groups (Table 1). For the MAS strategies the requirements to be eligible for selection were relaxed (outlined later). Selection indices were developed to calculate the accuracy of evaluation (variance of EBV) for the selection candidates, for a trait or index with heritability of 0.25 and repeatability of 0.6. The information sources in the base selection index were dam (two lactations), paternal half-sibs (85), paternal grand half-sibs (1000), maternal grand half-sibs (1000), paternal granddam performance (four lactations) and maternal granddam performance (four lactations). Selection indices for female selection candidates included lactation information (0–6 lactations) on the animal itself, and for the male selection candidates information from female progeny (0 or 85) was included. The additive genetic variance was partitioned between unmarked additive polygenic variation (which will be referred to as polygenic variance) and variation because of the QTL in linkage disequilibrium (referred to as marked genetic variance). The marked genetic variance was not assumed to be one QTL but many QTL with many alleles. The effects for each loci were assumed to range from large to small in accordance to the genetic mode described by Shrimpton and Robertson (1988). The marked genetic variance was included in the above selection index as an information source only for the selection candidate(s) in the selection index. The molecular marker data for the relatives provides no additional information for the selection candidate(s) in estimating genetic merit for the marked genetic variance. However, the data does allow more accurate estimation of the relatives and indirectly the breeding value

53

of the selection candidate (Lande and Thompson, 1990). In this study the marker data was not included as an information source for the relatives as the effect on accuracy of the selection candidate’s estimated breeding value is minor. Mean and variance of EBV was calculated for males and females for each age group (0–7 years). Truncation selection was undertaken across age groups eligible for selection (Ducrocq and Quaas, 1988). Based on normal distribution theory, the selected fraction and the standardized selection differential were calculated for each age group eligible for selection. Selection differentials were adjusted to account for finite population size using the approximation of Burrows (1972); (1 2 p) i f 5 i ` 2 ]]]]] (2i ` p(Ntot 1 1)

(1)

where i f is the finite selection intensity, i ` is the infinite selection intensity, p is the fraction selected from the number of individuals available for selection (Ntot ) in that age group. The effect of selection on the (co)variances of the information sources of the selected animals was calculated using a generalization of the Cochran (1951) formula.

s *jk 5 sjk (1 2 r 2TI i ` (i ` 2 t))

(2)

where s jk , and s *jk are the covariance of j and k before and after selection, respectively (when j5k it is the variance of j), r 2TI is the squared correlation between index and objective for the selected animals and t is the truncation point. Selection was undertaken for 50 years to ensure the breeding program reached equilibrium. Equilibrium rates of genetic gain are reported. Reduction in genetic variance through selection (Eq. (2)) (Bulmer, 1971), was modeled over the 50 years. The marked genetic variance as a proportion of the additive genetic variation was maintained at the same level over all years. Inbreeding was ignored in the model.

2.2. Marker assisted selection strategies Three different breeding schemes with varying proportions of genetic variance explained by the markers (0 to 100%) were investigated. The control

54

R. J. Spelman et al. / Livestock Production Science 59 (1999) 51 – 60

for each of the MAS schemes was when marked genetic variance was 0% for that scheme. All the genetic responses are in terms of annual gains in genetic standard deviations in the base population and percentage gains are relative to the appropriate control.

2.2.1. Base breeding program with marker information ( BMARK) The breeding scheme is the same as that described as the basebreeding program but with the additional information of markers. Marker information was exploited for all selection paths. 2.2.2. Young unproven bulls selected for the bull to bull path ( YBULL) Bulls selected for the bull to bull pathway could be selected either as 1 year olds (pedigree and marker information) or as 5 year olds (pedigree, progeny and marker information). Bulls that were selected as yearlings were also available for selection as 5 year olds. For the bull to cow selection path, only bulls with progeny records were eligible for selection. Marker information was used on all selection paths. 2.2.3. Open scheme ( OPEN) Bulls were eligible for selection from 1 year of age to 7 years of age for both bull to cow and bull to bull selection paths. Bulls could be selected in more than 1 year. Selection of young bulls for the bull to cow pathway is limited by semen production constraints. In the simulated breeding scheme it is assumed that all breeding occurs in a 3-month period (New Zealand dairy production system). Mature bulls ($3 years) are assumed to produce an average of 200 000 doses of semen over this period and yearling bulls 30 000 doses and 2-year-old bulls 150 000 doses (D. Hemara, personal communication). The truncation procedure applied across bull ages accounted for the lower semen capabilities of the younger bulls for the bull to cow path. Age constraints on the cow to bull path were relaxed to allow for selection from calves (5 months of age) and older animals. Reproduction from calves is dependent on techniques such as in vitro fertilization. Replacements on the cow to cow path are as in the

base scheme. Marker information is used on all selection paths. For MAS schemes (b) and (c) the age groups were divided into subgroups that reflected the amount of pedigree information. For instance in the open MAS scheme a young bull may be sired by a bull that was selected at 1 year of age through to a bull that was selected on progeny information. Likewise the number of lactations of dams of the young bulls may vary from none through to six. With differing amounts of pedigree information the accuracy of selection (variance of EBV) will differ and was accounted for. The amount of genotyping is similar for all three scenarios as it is assumed that all progeny tested sires and dams selected for the cow to bull pathway are genotyped. As no selection is undertaken on the cow to cow path, genotyping is not needed for the majority of cows. It is assumed that the majority of the QTL information will be collected before marker assisted selection begins, and there will be continued identification of new variation within marker haplotypes and new QTL during marker assisted selection. The contribution of each selection path to the increase in genetic gain, compared with the situation when there was no marked genetic variance, was calculated. This was determined by only including the marker information on the selection path of interest (e.g. bull to cow), and then comparing the genetic response to that when marker information was used on all selection paths.

3. Results The rate of genetic gain in the base breeding program with no marked genetic variance was 0.258 s G (Table 2). The rate of genetic gain increased as the proportion of marked genetic variance increased in the BMARK scheme (Table 2). The maximum percentage increase was 24% (0.26 to 0.32 s G ) when all of the genetic variance was marked (Table 2). Of the increase in genetic gain 80 to 95% was from the cow to bull path. The increase in genetic gain for each unit increase in marked genetic variance was approximately linear up to 50% marked genetic variance and then it increased in an exponential

R. J. Spelman et al. / Livestock Production Science 59 (1999) 51 – 60 Table 2 Rates of genetic gain in the BMARK a breeding program with different proportions of genetic variance explained by markers Marked genetic variance

Genetic gain b (s G / year)

Percent increase c

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

0.258 0.263 0.268 0.273 0.278 0.283 0.288 0.295 0.301 0.309 0.320

– 1.82 3.67 5.56 7.49 9.52 11.66 13.99 16.62 19.78 24.05

a

Sires must have a progeny test proof to be eligible for selection and cows must be 1 year of age. b Genetic standard deviation in the base population. c Compared to no marked genetic variance.

manner (Fig. 1). The greater rate of improvement in genetic gain for the larger proportions of marked genetic variance is due to the accuracy of evaluation increasing more per unit marked genetic variance

55

than at lower proportions of marked genetic variance, especially for sire evaluation as most of the genetic variance was explained through the progeny test of 85 daughters (Fig. 1). The marked genetic variance had little effect on the age of animals selected: the largest decrease in generation interval being 0.05 years for the cow to bull path over the range of 0 to 100% marked genetic variance. The rate of genetic gain for the breeding scheme with young bulls being used for the bull to bull path (YBULL) for 0% marked genetic variance was approximately the same as that when only proven bulls were used for that selection path (Table 2 and Table 3). The proportion of young bulls selected for this path was 0.24 resulting in a generation interval of 5.04 years compared to 6.28 years for the proven bull scheme. The rate of genetic improvement in the YBULL scheme increased by 52% when all of the genetic variance was marked. A linear increase in genetic gain for extra unit of marked genetic variance was observed (|0.0013 s G / % marked genetic variance). As the percentage of marked genetic variance moved

Fig. 1. Incremental increase in genetic gain (bars) and accuracy of sire evaluation (line) for each additional percent of marked genetic variance.

R. J. Spelman et al. / Livestock Production Science 59 (1999) 51 – 60

56

Table 3 Rates of genetic gain in the YBULL a breeding program with different proportions of genetic variance explained by markers

Table 4 Rates of genetic gain in the OPEN a breeding program with different proportions of genetic variance explained by markers

Marked genetic variance

Genetic gain b (s G / year)

Percent increase c

Percent young bulls d

Marked genetic variance

Genetic gain b (s G / year)

Percent increase c

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

0.260 0.271 0.284 0.300 0.313 0.326 0.339 0.352 0.365 0.379 0.395

– 4.49 9.42 15.46 20.45 25.45 30.48 35.55 40.72 46.12 52.09

24.1 56.0 74.6 86.0 92.3 95.8 97.7 98.8 99.2 99.5 99.6

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

0.282 0.301 0.330 0.360 0.406 0.437 0.467 0.497 0.525 0.551 0.577

– 6.73 17.28 27.85 44.09 55.16 65.88 76.33 86.23 95.69 104.71

a

a

from zero to 10% the proportion of bulls selected as young bulls for the bull to bull path increased from 24% to 56% (Table 3) and surpassed 90% when marked genetic variance reached 36%. Therefore the rate of genetic gain in a breeding scheme where only young bulls are eligible for selection for the bull to bull path is nearly equal to that of young and old bulls when marked genetic variance is more than 30%. The contribution to improvement in genetic gain from each of the three selection paths was, on average (over all proportions of marked variance), |45% from the cow to bull path, 5% from the bull to cow path and 50% from the bull to bull path in the YBULL scheme. The rate of genetic gain with the OPEN scheme with no marked genetic variation was 0.28 s G (Table 4). The proportion of unproven bulls selected for the bull to cow pathway was 0.67 and 0.37 for the bull to bull pathway, when no genetic variation was marked. Seventy percent of the cows selected as bull mothers did not have lactation information. The average generation interval in the OPEN scheme for the bull to cow path was 3.67 years and 4.79 years for the bull to bull path and 2.30 years for the cow to

bull path when there was no marked genetic variance. The rate of genetic gain for the OPEN scheme when all of the genetic variance was marked was 0.57 s G , which is an increase of 105% over the scheme with no marked genetic variance. The rate of increase in genetic gain when marked genetic variance increased from 0 to 10% was 7%. The largest percentage increase in genetic gain was when marked genetic variance increased from 30 to 40%. On average, an increase of 1% marked genetic variance resulted in an increase of |1% in genetic gain. Sixty percent of the increase in genetic gain originated from the cow to bull selection path when up to 40% marked genetic variance was marked, and reduced to 50% for higher levels of marked genetic variance. The marker information benefited the bull to bull pathway more than the bull to cow selection path with some 30% of the extra genetic response from the bull to bull path and 20% from the bull to cow path. When all of the genetic variance was marked the proportion of unproven bulls for the bull to cow pathway and for the bull to bull pathway was more than 0.99, as was the proportion of cows selected without lactation information for the cow to bull selection path. The average generation interval for the bull to cow path was 2.34 years and 2.18 years

Sires for the bull to bull selection path are eligible for selection as either 2 year olds (unproven) or as 5 year olds (proven). Sires must have a progeny test proof to be eligible for selection for the bull to cow path and cows must be 1 year of age. b Genetic standard deviation in the base population. c Compared to no marked genetic variance. d Percentage of sires selected for the bull to bull selection path that are 2 years of age.

Sires are eligible for selection for both sire selection paths from 1 year of age and cows are eligible for selection as calves and older age groups. b Genetic standard deviation in the base population. c Compared to no marked genetic variance.

R. J. Spelman et al. / Livestock Production Science 59 (1999) 51 – 60

for the bull to bull path and 1.6 years for the cow to bull path when all of the genetic variance was marked.

4. Discussion

4.1. Simulation model It has been shown in MAS simulation studies, with a genetic model comprising of polygenic and a small number of marked loci, that selection alters the allele frequency at the marked loci, which decreases the variation at the loci as they near fixation (Meuwissen and Goddard, 1996; Ruane and Colleau, 1996). In this study it was assumed that the variance of the marked QTL was not affected by allele frequency changes, but only reduced by the negative covariance between loci generated by selection (Bulmer, 1971). Therefore the marked genetic variance as a proportion of the additive genetic variance was static over years. This was assumed as this study concentrated on the genetic response when large proportions of marked genetic variance were identified. In this situation it is likely that there will be many loci marked and changes in allele frequency at each locus will be small. In addition, it would be inconsistent to assume there are only allele frequency changes when the genetic variance is marked and ignore changes when the genetic variance is treated as polygenic. The fixed proportion of additive genetic variance being marked may be realistic as a result of new QTL being identified over time as new biochemical pathways become the rate limiting step. Also new genetic variation may be found within certain haplotypes as one gets more genotypic and phenotypic information. The genetic model assumed in this study was that there are many QTL, and the QTL effects range from a small number of loci with large effects to many loci with smaller effects. It is acknowledged by the authors that the genetic model simulated is rather simplistic but it was chosen in this manner to enable easy computation for the deterministic model. The model will be robust in the authors opinion for marked genetic variances of up to 50%. For marked variances above 50% the assumptions of no reduction in genetic variance through allele changes and a

57

fixed marked genetic variance and no linkage between QTL are less likely to hold. In addition it should be noted that the reported genetic responses are likely to be the upper bound for the given breeding programmes. In this study the percentage of marked genetic variance was varied from 0 to 100%. Although the true underlying genetic model is unknown (e.g. number of loci, distribution of effects, and interaction of loci) it is unlikely that all of the genetic variance will be able to be marked as experimental power will be too low to detect loci with small effects, and with epistatic effects. The proportion of genetic variance that the loci with small effects constitute is unknown. However, with experimental techniques such as selective DNA pooling up to 50–75% of the genetic variance for one trait has been identified (Soller, personal communication). Therefore it can be expected that large proportions (up to 50%) of genetic variance should be identified in dairy cattle populations in future. The accomplishment of identifying loci that are in linkage disequilibrium with marker loci (haplotypes) will be challenging for molecular and quantitative geneticists. Risch and Merikangas (1996) report that for a complex disease trait in humans, the statistical power of linkage disequilibrium mapping is greater than that of linkage analysis. These authors see the primary limitation of genome-wide association tests as not a statistical one but a technological one. The technological limitations that they saw, were the identification of a large number of polymorphisms and the testing of these polymorphisms on a large number of individuals. These technological limitations may be overcome with the development of single nucleotide polymorphisms (SNPs) as biallelic markers. Kruglyak (1997) reported that the SNPs are highly abundant with classic estimates of 1 per 1000 base pairs or more than 3 million in the genome. Kruglyak (1997) also reported that the use of these highly abundant markers with non-gel based assays (DNA chips) is promising. This technology has the potential to enable genome scans for linkage disequilibrium and association studies. However, Baret and Hill (1997) conclude that the application of linkage disequilibrium mapping to livestock will be limited due to insufficient knowledge of the genetic history of the population and the operation of

58

R. J. Spelman et al. / Livestock Production Science 59 (1999) 51 – 60

disruptive factors such as selection and drift. These authors see the application of linkage disequilibrium to livestock being limited to discrete traits in specific populations (isolated populations or populations stemming from an admixture event). Further research in this area is needed. Many alternative breeding schemes could have been investigated, but of the three chosen, BMARK represents a traditional breeding, and the OPEN scheme represents the other end of the spectrum, and the YBULL scheme an intermediary scheme. Other MAS strategies such as pre-selection of bulls entering the progeny test were not investigated.

4.2. Genetic response The levels of genetic response that are presented in this study are not what could be achieved today but possibly in 10 years time. If the assumptions used in this study of large proportions of marked genetic variation in linkage disequilibrium with markers are realised, marker assisted selection has the potential to have a considerable impact on dairy cattle breeding schemes. Meuwissen and van Arendonk (1992) reported similar increases in rates of genetic gain for a progeny test scheme that is comparable to the BMARK scheme. The rates of improvement in genetic gain of |4% for the progeny-test-based MAS scheme (BMARK) when 20% of the genetic variance is explained is similar to gains outlined in other studies on progeny-test-based breeding programs (Spelman and Garrick, 1997, 1998; Mackinnon and Georges, 1998). For the OPEN and YBULL schemes the most comparable study is that of Meuwissen and Goddard (1996) where they simulated a closed nucleus breeding scheme. Those authors reported an increase in genetic gain of 38% when 33% of the genetic variance was marked, which is a similar result to that of the OPEN scheme. Rates of genetic gain presented in this study are at equilibrium, which are only reached after 20–30 years of selection. When changing the breeding program the genetic responses in the immediate years can fluctuate quite dramatically (Ducrocq and Quaas, 1988). If a MAS breeding scheme is implemented that is quite different from the current scheme the genetic response in years immediately

after implementation should be investigated, because it will be of importance to the breeding company in terms of retaining and increasing market share until equilibrium response is reached. Another aspect of the breeding scheme to be investigated would be the variance of genetic response. Meuwissen (1991) reported that breeding schemes with the shortest generation intervals had the highest variance of response. The increase in variance of response will be less than that reported by Meuwissen (1991), as marker information increases the accuracy of the genetic merit estimate for the younger animals compared to the situation of no marker information as in Meuwissen (1991). Marker assisted selection also has the additional risk factor of errors in estimation of location and size of the marked genetic variance. The decision on implementation of a genetically superior breeding scheme with larger variance of genetic response will depend on the degree of risk aversion. In all three of the MAS schemes it was assumed that 140 bulls were progeny tested each year. For the open scheme when 40% of the genetic variance was marked 95% of the bulls selected for the two sire paths were unproven bulls. In this situation progeny testing the bulls adds little to the genetic response and only adds expense to the breeding scheme. However, if progeny testing is required as a tool to market bulls, then one could pre-select the young bulls to progeny test when marked variance is $40% (also possible at lower levels), which would reduce the cost. For the OPEN scheme without progeny testing the genetic response is near equal to that with progeny testing when marked variance is 20% and greater. In addition, costs for the breeding scheme with the young bull breeding scheme will be reduced as no progeny testing is undertaken. In the BMARK scheme, marked genetic information added little to the bull selection paths as the progeny test on 85 daughters explained most of the genetic variance. If an organisation was not willing to change its breeding program away from proven bulls another strategy may be to undertake the progeny test on fewer daughters. Assuming the total number of daughters in the progeny test program is fixed, more sires could be progeny tested. For example, for marked genetic variance of 20% and 198 bulls progeny tested on 60 daughters, genetic

R. J. Spelman et al. / Livestock Production Science 59 (1999) 51 – 60

gain increased by 6.7% over the base situation of 140 bulls progeny tested on 85 daughters and no marked genetic variance. This increase is nearly double the percentage increase for 20% marked genetic variance and 140 bulls and 85 daughters (Table 2). However, the genetic advantage may not be an economic advantage once the costs of producing and feeding / housing the extra bulls is accounted for (Meuwissen, 1997). Relaxing the constraint on age of selection for the bull to bull path, resulted in a breeding scheme (YBULL) that was able to benefit more from the identification of QTL. On average, the percent increase in genetic gain was double for the YBULL scheme compared to the BMARK scheme (Table 2 and Table 3). When the age of selection was relaxed further on the bull to cow and cow to bull pathways (OPEN) the marked genetic information was utilised more efficiently again. On average, the percentage increase in genetic gain for the OPEN scheme was double that of the YBULL scheme (Table 3 and Table 4). Therefore, to optimally use the marked genetic information, breeding programs will have to move away from the progeny test system and select animals without phenotypic and progeny information. Non-acceptance of young bulls for the bull to cow path by the semen users could hinder implementation of a breeding program such as the OPEN scheme. This would not be the case for the YBULL scheme as decisions for the bull to bull path are made by the breeding organisations and the sires selected by the semen users will still have approximately the same reliability as the BMARK scheme. There is increasing interest in the use of young bulls with 0% marked genetic variance, and as shown in this study, the benefit of using young bulls improves as the rate of genetic gain increases. This study shows that with medium to large proportion of genetic variance identified and being in linkage disequilibrium with marker loci, MAS can substantially increase the rate of genetic gain. To utilise the marker information breeding companies will have to alter their breeding schemes from the traditional progeny test system to schemes selecting animals without lactation and progeny information. However this may not be appropriate at low proportions of marked genetic variance as the cost of

59

altering the breeding scheme may be greater than the benefits.

References Baret, P.V., Hill, W.G., 1997. Gametic disequilibrium mapping: potential applications in livestock. Anim. Breed. Abs. 65 (5), 309–318. Bulmer, M.G., 1971. The effect of selection on genetic variability. Am. Natural. 165 (943), 201–211. Burrows, P.M., 1972. Expected selection differentials for directional selection. Biometrics 28, 1091–1100. Charlier, C., Farnir, F., Berzi, P., Vanmanshoven, P., Brouwers, B., et al., 1996. Identity-by-descent mapping of recessive traits in livestock: application to map the Bovine Syndactyly locus to chromosome 15. Genome Res. 6, 580–589. Cochran, W.G., 1951. Improvement by means of selection. In: Neyman, J. (Ed.), Proceedings of the 2nd Berkley Symposium on Mathematical and Statistical Problems, pp. 449–470. Darvasi, A., Soller, M., 1994. Selective DNA pooling for determination of linkage between a molecular marker and a quantitative trait locus. Genetics 138, 1365–1373. Ducrocq, V., Quaas, R.L., 1988. Prediction of genetic response to truncation selection across generations. J. Dairy Sci. 71, 2543– 2553. Georges, M., Nielsen, D., Mackinnon, M., Mishra, A., Okimoto, R., et al., 1995. Mapping quantitative trait loci controlling milk production in dairy cattle by exploiting progeny testing. Genetics 139, 907–920. Kruglyak, L., 1997. The use of a genetic map of biallelic markers in linkage studies. Nature Genet. 17, 21–24. Lande, R., Thompson, R., 1990. Efficiency of marker assisted selection in the improvement of quantitative traits. Genetics 124, 743–756. Lipkin, E., Mosig, M.O., Darvasi, A., Ezra, E., Shalom, A., et al., 1998. Mapping loci controlling milk protein percentage in dairy cattle by means of selective milk DNA pooling using dinucleotide microsatellite markers. Genetics (in press). Mackinnon, M.J., Georges, M., 1998. A bottom up approach to marker assisted selection. Livestock Prod. Sci. (in press). Meuwissen, T.H.E., 1991. Expectation and variance of genetic gain in open and closed nucleus and progeny testing schemes. Anim. Prod. 53, 133–141. Meuwissen, T.H.E., 1997. Optimization of progeny tests with prior information on young bulls. Livestock Prod. Sci. 52 (1), 57–68. Meuwissen, T.H.E., Goddard, M.E., 1996. The use of marker haplotypes in animal breeding schemes. Genet. Sel. Evol. 28, 161–176. Meuwissen, T.H.E., van Arendonk, M.E., 1992. Potential improvements in rate of genetic gain from marker assisted selection in dairy cattle breeding schemes. J. Dairy Sci. 75, 1651–1659. Risch, N., Merikangas, K., 1996. The future of genetic diseases of complex human diseases. Science 273, 1516–1517.

60

R. J. Spelman et al. / Livestock Production Science 59 (1999) 51 – 60

Ruane, J., Colleau, J.J., 1996. Marker-assisted selection for a sex-limited character in a nucleus breeding population. J. Dairy Sci. 79, 1666–1678. Shrimpton, A.E., Robertson, A., 1988. The isolation of polygenic factors controlling bristle score in Drosophila melanogaster. II. Distribution of third chromosome bristle effects within chromosome sections. Genetics 118, 445–459. Smith, C., 1967. Improvement of metric traits through specific genetic loci. Anim. Prod. 9, 349–358. Smith, C., Smith, D.B., 1993. The need for close linkages in marker assisted selection for economic merit in livestock. Anim. Breed. Abs. 61 (4), 197–204.

Spelman, R.J., Garrick, D.J., 1997. Utilisation of marker assisted selection in a commercial dairy cow population. Livestock Prod. Sci. 47, 139–147. Spelman, R.J., Garrick, D.J., 1998. Genetic and economic responses for within-family marker-assisted selection in dairy cattle breeding schemes. J. Dairy Sci. (in press). Spelman, R.J., Lopez-Villalobos, N., Garrick, D.J., 1998. Experimental designs for quantitative trait loci detection in the New Zealand dairy industry. Proc. NZ Soc. Anim. Prod. 58, 6–9.